CN110175272A - One kind realizing the convergent control method of work order and control device based on feature modeling - Google Patents
One kind realizing the convergent control method of work order and control device based on feature modeling Download PDFInfo
- Publication number
- CN110175272A CN110175272A CN201910423511.7A CN201910423511A CN110175272A CN 110175272 A CN110175272 A CN 110175272A CN 201910423511 A CN201910423511 A CN 201910423511A CN 110175272 A CN110175272 A CN 110175272A
- Authority
- CN
- China
- Prior art keywords
- work order
- information
- work
- feature
- characteristic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/906—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
- G06Q10/103—Workflow collaboration or project management
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Economics (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides one kind to realize the convergent control method of work order based on feature modeling, it is based on dynamic acquisition data, summarizes rule modeling, and work order data application is realized to work order restrains in modeling, include the following steps: that work order determines one or more characteristic informations in one or more work orders to a. based on one or more；B. one or more characteristic informations characteristic model corresponding with the characteristic information in one or more work orders is matched one by one, obtains one or more convergence work orders and/or one or more not converged work orders.The present invention is powerful, easy-to-use, facilitates that fast, convergence is good, has high commercial value.
Description
Technical field
The invention belongs to artificial intelligence application fields, and in particular, to one kind realizes that work order is convergent based on feature modeling
Control method and control device.
Background technique
Maintenance work is had been manually done by operation maintenance personnel, with Internet service Rapid Expansion, human cost
The epoch of height enterprise, people's meat O&M are difficult to maintain day-to-day operation.Then pass through the foot of creation can be automatically triggered predefined rule
This realizes automation O&M, it is considered to be one kind is based on industry field knowledge to execute common, repeated maintenance work
With the expert system of O&M scenarios domain knowledge.It is flat as core building O&M automation using layout operation mode to automate O&M
Platform can provide the functions such as configuration change, task inspection, script execution control, Custom Workflow for client；Cover inspection, text
The O&M scenarios such as part distribution, Backup and Restore, SQL operation, and expandability is provided；Automate the configuration of operational system server end
Batch job strategically dispatches each script work, realizes the automatic operation of Business Stream；Console is the operation of each administrative staff
Platform, realize role, object, operation classification decentralized management.Unitized device model is constructed on the basis of CMDB, realizes resource
Automation with service is delivered.
IT O&M is developed so far from birth, undergoes artificial O&M to current automation O&M, has been able to gradually adapt to
The business of complexity, diversified user demand, the IT application constantly extended, to ensure that IT services flexibly convenient, safety and stability, from
Dynamicization O&M is the demand of starting point as manual operation is replaced, and is just widely studied and applied.O&M automation is one group and incites somebody to action
Static device structure is converted into the strategy responded according to IT demand for services dynamic elasticity, and purpose is exactly to realize the matter of IT O&M
Amount, reduces cost.Automation O&M mainly to create the script of predefined rule that can be automatically triggered, come execute it is common,
The maintenance work of repeatability realizes automation O&M.But the problem of not being predefined once occurs, it is necessary to artificially go to feel
Know, check, position, the manpower that this O&M service needs to put into is quite huge.Because shortage of manpower is positioned because checking, will lead
It causes service to hang up for a long time, cannot achieve the stable operation of system.This deficiency is increasingly prominent, this is also that intelligent O&M brings hair
Open up opportunity.
Intelligent O&M is advocated by machine learning algorithm independent of artificial specified rule automatically from magnanimity operation/maintenance data
Constantly learn in (the artificial treatment log including event itself and operation maintenance personnel), constantly refines and summarize rule.I.e.
It is to increase the brain based on machine learning on the basis of automating O&M, Command Monitoring System acquires brain decision
Required data make analysis, decision, and command automation script goes to execute the decision of brain, to reach operational system
Overall goals.
And currently, there is no a kind of specific way that can effectively solve the problem that the above problem more particularly to a kind of bases in the market
The convergent control method of work order and control device are realized in feature modeling.
Summary of the invention
For technological deficiency of the existing technology, the object of the present invention is to provide one kind to realize work order based on feature modeling
Convergent control method and control device, according to an aspect of the invention, there is provided a kind of realize work order based on feature modeling
Convergent control method based on dynamic acquisition data, summarizes rule modeling, and work order data application is realized work in modeling
Single convergence, includes the following steps:
A. work order determines one or more characteristic informations in one or more of work orders based on one or more；
B. one by one by one or more characteristic informations spy corresponding with the characteristic information in one or more work orders
Sign model is matched, and one or more convergence work orders and/or one or more not converged work orders are obtained.
It preferably, further include step c after the step b: matching and one or more convergence work orders and/or one
Or the work order information that multiple not converged work orders are adaptable, and the work order information is subjected to Dynamic Display.
Preferably, in the step a, the characteristic information is included at least:
Work order classification information；
Information extraction information；
Work order coverage information；And
Root cause analysis information.
Preferably, the step b includes the following steps:
B1. the work order classification information based on the work order judges whether to converge to fisrt feature model group, if so, holding
Row step b2；
B2. the information extraction information in fisrt feature model group based on the work order judges whether to converge to
Two characteristic model groups, if so, executing b3；
B3. the work order coverage information in second feature model group based on the work order judges whether to converge to
Three characteristic model groups, if so, executing b4；
B4. the root cause analysis information in third feature model group based on the work order judges whether to converge to
Four characteristic model groups.
Preferably, the foundation of the characteristic model will obtain in the following way:
I: one or more originals are obtained in one or more data acquisition platforms based on one or more data acquisition modes
Beginning work order information；
Ii: pre-processing one or more of original work order informations, determines one or more pretreatment work order letters
Breath；
Iii: carrying out feature extraction to one or more of pretreatment work order informations, determines one or more work order feature
Information；
Iv: being based on one or more of work order Feature information modelings, determines that the one or more and work order feature is believed
The matched characteristic model of manner of breathing.
Preferably, the data acquisition modes include at least such as any one of under type or appoint a variety of:
SQL export；
Third party API export；Or
Crawler acquisition.
Preferably, the data acquisition platform includes at least such as any one of lower platform or appoints a variety of:
O&M worksheet processing system；
Configuration management database；
Application performance monitoring；
Banner system；Or
Unified log platform.
Preferably, the original work order information includes at least any one of following information or appoints a variety of:
Work order；
CMDB；Or
APM。
Preferably, the pretreatment includes attachment acquisition, data screening and data encoding.
Preferably, the convergence work order and the not converged work order include at least following two aspects:
One or more failures are in one or more one or more work orders for monitoring and generating under dimension；
The appearance repeatedly that the alarm of one or more caused by one or more reasons phenomenon is inscribed in one or more.
According to another aspect of the present invention, it provides one kind and the convergent control device of work order is realized based on feature modeling,
Include:
First determining device: work order determines one or more features letter in one or more work orders based on one or more
Breath；
First acquisition device: one by one by the one or more characteristic informations and the characteristic information in one or more work orders
Corresponding characteristic model is matched, and one or more convergence work orders and/or one or more not converged work orders are obtained.
First processing unit: matching is adapted with one or more convergence work orders and/or one or more not converged work orders
Work order information, and by the work order information carry out Dynamic Display.
Preferably, first acquisition device includes the following:
First judgment means: the work order classification information based on the work order judges whether to converge to fisrt feature Model Group
Group；
Second judgment means: in fisrt feature model group based on the work order information extraction information judgement be
It is no to converge to second feature model group；
Third judgment means: in second feature model group based on the work order work order coverage information judgement be
It is no to converge to third feature model group；
4th judgment means: in third feature model group based on the work order root cause analysis information judgement be
It is no to converge to fourth feature model group.
Preferably, further includes:
Second acquisition device: one is obtained in one or more data acquisition platforms based on one or more data acquisition modes
A or multiple original work order informations；
Second determining device: pre-processing one or more of original work order informations, determines one or more pre-
Handle work order information；
Third determining device: feature extraction is carried out to one or more of pretreatment work order informations, determines one or more
A work order characteristic information；
4th determining device: being based on one or more of work order Feature information modelings, determine it is one or more with it is described
The characteristic model that work order characteristic information matches.
The present invention is based on dynamic acquisition data, rule modeling is summarized, work order determines one or more based on one or more
One or more characteristic informations in work order, one by one by one or more work orders one or more characteristic informations and the spy
The corresponding characteristic model of reference manner of breathing is matched, and one or more convergence work orders and/or one or more not converged works are obtained
It is single, i.e., work order data application is realized to work order restrains in modeling, the present invention is powerful, easy-to-use, it is fast to facilitate, convergence
Property it is good, have high commercial value.
Detailed description of the invention
Upon reading the detailed description of non-limiting embodiments with reference to the following drawings, other feature of the invention,
Objects and advantages will become more apparent upon:
Fig. 1 shows a specific embodiment of the invention, and one kind realizing the convergent controlling party of work order based on feature modeling
The idiographic flow schematic diagram of method；
Fig. 2 shows the first embodiment of the present invention, one by one by one or more features in one or more work orders
Information characteristic model corresponding with the characteristic information is matched, obtain one or more convergence work orders and/or one or
The idiographic flow schematic diagram of multiple not converged work orders；
Fig. 3 shows the second embodiment of the present invention, establishes the idiographic flow schematic diagram of the characteristic model；
Fig. 4 shows another embodiment of the present invention, and one kind realizing the convergent control of work order based on feature modeling
The module connection diagram of device processed；And
Fig. 5 shows the third embodiment of the present invention, and one kind realizing the convergent control device of work order based on feature modeling
Module connection diagram.
Specific embodiment
In order to preferably technical solution of the present invention be made clearly to show, the present invention is made into one with reference to the accompanying drawing
Walk explanation.
Fig. 1 shows a specific embodiment of the invention, and one kind realizing the convergent controlling party of work order based on feature modeling
The idiographic flow schematic diagram of method, specifically, the present invention provides one kind to realize the convergent control method of work order based on feature modeling,
It is based on dynamic acquisition data, summarizes rule modeling, and work order data application is realized that work order restrains in modeling, including as follows
Step:
Firstly, enter step S101, based on one or more work order determine one in one or more of work orders or
Multiple characteristic informations, it will be appreciated by those skilled in the art that in the present invention, its work can be determined based on one or more work order
Characteristic information in list, the work order are preferably showed in table form, extract text information in the work order into
Row classification, the as described characteristic information, the characteristic information include at least work order classification information；Information extraction information；Work order is received
Hold back information and root cause analysis information.
Further, work order classification information is and will alert work order to classify and dated information by fault type, is subsequent
The important foundation of processing.Work order classification substantially short essay present treatment, it will be appreciated by those skilled in the art that the present invention, which uses, is based on depth
The text-processing model of neural network LSTM.With it is traditional based on the method for keyword and regular expression compared with, classification is accurate
Property promoted from 70%-80% to 95%-97%.
Long Short Term network-LSTM-- is a kind of special RNN type, can learn long-term Dependency Specification.
LSTM is proposed by Hochreiter&Schmidhuber (1997), and is improved and promoted by Alex Graves.Very much
Problem, LSTM obtains quite huge success, and is widely used in a variety of applications.LSTM by design deliberately (cell state,
Forget door, Memory-Gate, out gate) avoid long-term Dependence Problem.Remembeing long-term information in practice is the default row of LSTM
For, and non-required just obtainable ability of paying a high price.The present invention, which is namely based on LSTM, realizes a work order classification mould
Type, to determine the operation in subsequent step by work order classification information, these will be described below in specific embodiment make into
The description of one step, it will not be described here.
Further, it will be appreciated by those skilled in the art that in the present invention, all characteristic informations will be used in character modules
Work order convergence is carried out in type, and characteristic model is to meet current operation management according to what a large amount of original work order information extracted
Model, and current work order information is updated to in the corresponding model of the work order, work order can be restrained, and in the present invention
Characteristic information is in work order for convergent corresponding informance, further includes information in addition to the work order classification information mentioned in aforementioned
Information, work order coverage information and root cause analysis information are extracted, this is because it corresponds respectively to its corresponding model, these will
It is further described in the examples described below, it will not be described here.
Then, enter step S102, one by one by one or more work orders one or more characteristic informations and the spy
The corresponding characteristic model of reference manner of breathing is matched, and one or more convergence work orders and/or one or more not converged works are obtained
Single, in such embodiments, the present invention is based on step S101 to get corresponding in one or more work orders one or more
A characteristic information, and by one or more features information and the characteristic information phase corresponding in one or more of work orders
Corresponding characteristic model is matched, and characteristic model, the specific reality that the building of the characteristic model will be described below are referred here to
It applies in mode and is further described through.Herein, the quantity of the characteristic model preferably with the quantity phase of the characteristic information
Matching, can for 4,5 it is even more, if the characteristic information matches with the characteristic model, carry out work order receipts
It holds back, if it does not match, not restraining.
It further include step S103 after executing the step S102 as a preferred embodiment of the invention, matching
The adaptable work order information with one or more convergence work orders and/or one or more not converged work orders, and the work order is believed
Breath carries out Dynamic Display.It will be appreciated by those skilled in the art that the present invention is based on the alarms of machine learning techniques research and development to restrain model simultaneously
It is embedded in current alarming processing process, completes to alert the convergence convergence of (including history), root because of positioning, with automation operation platform
Link realizes the efficient linking that the association analysis after merging, O&M automatically process, and combines big data visualization technique by seeing
Plate form vividly shows output.
It will be appreciated by those skilled in the art that Fig. 1, which basically illustrates one kind, realizes the convergent controlling party of work order based on feature modeling
Method, according to the definition of Gartner, it is exactly that O&M has that AIOps, which is Algorithmic IT Operations (intelligent O&M),
The ability of machine learning and algorithm is based on existing operation/maintenance data (log, monitoring information, application message etc.), passes through engineering
The mode of habit can not solve further to solve the problems, such as to automate O&M.AIOps is based on automation O&M, by engineering
Habit, AI and O&M combine well, are that the development of O&M is inevitable, are the next developing stage for automating O&M.
The global deployment rate of Gartner correlation report prediction AIOps will increase to the 50% of the year two thousand twenty from 10% in 2017.It is applied
Industry further includes high-performance calculation, telecommunications, finance, electric power networks, Internet of Things, medical network and sets other than internet
The fields such as standby, aerospace, military equipment and network.The common application scenarios of AIOps include quality assurance, cost management and efficiency
Promoted etc..
Current each enterprise has all deployed corresponding monitoring alarm platform, takes effective prison to daily service operation situation
Control.With business high speed development, insurance information system quantity sustainable growth, transaction complexity is higher.When failure occurs, deployment
All kinds of monitor supervision platform concentrated alarms, high-volume warning message come tumbling through various channels, the continuous brush screen of mobile phone information, allow fortune
Dimension engineer has more visitors or business than one can attend to, and processing platform has overstock large quantities of fault warning work orders to be processed in the short time.Traditional fortune
Dimensional pattern arrangement 7*24 hours are on duty, once it alerts, it has to start with from single fault warning, while organization arrangement is multi-party
Personnel participate in fault location, reason retrospect.The pressure that fast quick-recovery business is normally carried out, allows O&M engineer's pressure multiplier, leads to
It crosses traditional O&M technology and improves solution timeliness as the bottleneck for being difficult to break through.
Fig. 2 shows the first embodiment of the present invention, one by one by one or more features in one or more work orders
Information characteristic model corresponding with the characteristic information is matched, obtain one or more convergence work orders and/or one or
The idiographic flow schematic diagram of multiple not converged work orders, includes the following steps:
Firstly, entering step S1021, the work order classification information based on the work order judges whether to converge to fisrt feature mould
Type group, it will be appreciated by those skilled in the art that the step S1021 is that judgment step basically illustrates four kinds of features in Fig. 2
Pattern die group, and in other examples, it can also be 3 groups or 5 groups, more specifically, the fisrt feature Model Group
Group can be with work order disaggregated model, specifically, the work order classification information based on the work order, is updated to the work order classification mould
In type, to judge whether the work order is same system etc., if it is not, then obtaining the not converged work order, and tie
Beam judgement, if so, carrying out work order convergence, and does and further restrains, that is, enter step S1022.
Then, S1022 is entered step, after executing the step S1021, the work order after step S1021 convergence is believed
Breath executes step S1022 again, i.e., the information extraction information judgement in fisrt feature model group based on the work order
Whether second feature model group is converged to, it will be appreciated by those skilled in the art that second feature model group is information extraction
Model, the information extraction information based on the work order, is updated in the information extraction model, to judge whether described
Whether work order is same timeslice, same place, region etc. information, if it is not, then obtaining the not converged work order, and is terminated
Judgement if so, carrying out work order convergence, and is done and is further restrained, that is, enters step S1023.
Subsequently, S1023 is entered step, the work order in second feature model group based on the work order restrains letter
Breath judges whether to converge to third feature model group, it will be appreciated by those skilled in the art that third feature model group is
Work order restrains model, and the work order coverage information based on the work order is updated in the work order convergence model, to judge
Whether whether the work order is same cluster information etc. information, if it is not, then obtaining the not converged work order, and terminates to sentence
It is disconnected, if so, carrying out work order convergence, and do and further restrain, that is, enters step S1024.
Finally, S1024 is entered step, based on the root cause analysis information of the work order in third feature model group
Judge whether to converge to fourth feature model group, it will be appreciated by those skilled in the art that fourth feature model group is root
Because of analysis model, the root cause analysis information based on the work order is updated in the root cause analysis model, so that judgement is
The no work order fault type etc. information, if it is not, then obtaining the not converged work order, and terminates to judge, if so, carrying out
Work order convergence.
Fig. 3 shows the second embodiment of the present invention, establishes the idiographic flow schematic diagram of the characteristic model, this field
Technical staff understands that Fig. 3 is used for the foundation of the matched characteristic model of characteristic model by showing in step s 102, specifically, packet
Include following steps:
Firstly, S201 is entered step, based on one or more data acquisition modes in one or more data acquisition platforms
One or more original work order informations are obtained, in such embodiments, the data acquisition modes are exported including at least SQL；
Third party API export or crawler acquisition.The data acquisition platform includes at least O&M worksheet processing system, configuration management data
Library, application performance monitoring, banner system or unified log platform.The original work order information include at least work order, CMDB or
Person APM.Data of the invention cover operation management system from hooligan more, including O&M worksheet processing system, configuration management database, answer
With performance monitoring, banner system, unified log platform etc..Data source include Oracle, MySQL, MongoDB,
ClickHouse etc..
Then, S202 is entered step, one or more of original work order informations are pre-processed, determines one or more
A pretreatment work order information, the pretreatment include attachment acquisition, data screening and data encoding.Part work order is because problem is retouched
It states too long, has been placed in attachment.Although these work order quantity are few, important.This system is in a manner of crawler from work
Single delivering system downloads accessory information, is added into related work order.Work order data include more than 80 a fields, are received mostly with work order
It is unrelated to hold back purpose.This system has screened event serial number, subevent serial number, event code, said system, problem description, work order life
At input of the fields as collective system such as time, work order source, affiliated management groups, and in work order classification and information extraction stage
The information such as fault type, IP are extracted from problem description, Cluster information is extracted from CMDB according to IP.For systematic name
Deng main to be converted to corresponding code using random coded scheme.
Subsequently, S203 is entered step, feature extraction is carried out to one or more of pretreatment work order informations, determines one
The selection of a or multiple work order characteristic informations, input feature vector is most important to modeling success or failure.The present invention uses professional knowledge and number
Feature is selected according to the mode combined is excavated.First by way of visiting business personnel, sample cases are collected, and may be with work
It is single to restrain relevant feature, the correlation of each feature with relevant variable is then calculated, suitable candidate feature is therefrom screened.
Pearson correlation coefficients and mutual information coefficient has mainly been used to measure univariate correlation in the present invention.Obtain candidate feature
Afterwards, according to these feature training patterns, according to the conspicuousness of the accuracy of model and recall rate assessment feature.
Finally, enter step S204, be based on one or more of work order Feature information modelings, determine it is one or more with
The characteristic model that the work order characteristic information matches.It will be appreciated by those skilled in the art that the characteristic model includes work order classification
Model, information extraction model, work order convergence model and root cause analysis model, we are in abovementioned steps S101 to work order
Disaggregated model is described, and in explanation below, we will to information extraction model, work order convergence model and root because
Analysis model is described in detail.
Further, information extraction model is mainly concerned with information extraction, and the main stream approach of information extraction includes based on certainly
The information extraction of right Language Processing, the information extraction based on ontology, the information extraction based on regular expression, and rise in recent years
The information extraction based on deep learning method.Decimation rule based on natural language processing method use filtering, part of speech and
Lexical semantic mark customizes decimation rule to establish the connection between phrase and Statement element by providing specific sample.?
That is text segmentation is marked each clause's ingredient at multiple clauses, then by the sentence analyzed and realization
Language rule mode is matched to obtain the content of clause.Rule can be worked out by manual type, can also be from artificial mark
Study obtains automatically in the corpus of note.
The decimation rule of method based on natural language processing is based primarily upon grammer limitation, semantic category, context and sentence
At grading information, ability to express is strong.But it does not have the structure feature using alarm work order itself during extracting, but
Effective decimation rule is obtained by a large amount of sample learning, user is needed largely to mark training sample, is only applicable to tie
The low free text of structure degree.Ontology is a kind of specific Formal Specification explanation of shared conceptual model.Based on this
The information extraction method of body mainly uses the thought of first mode, inputs as some of document to be processed and manual compiling applications
The ontology in field exports the information for structuring by extraction system.As long as the advantages of system is the application field created in advance
Ontology it is powerful enough, system can to the various webpages in a certain application field realize information extraction.The disadvantage is that system makes
With more troublesome, the ontology within the scope of a certain application field can only be created by the knowledge professional in the field.Based on deep learning
The information extraction mode of method is the enhancing of preceding method, can be improved the accuracy of preceding method, but there is also use fiber crops
It is tired, the disadvantages of needing user largely to mark the help of training sample and domain-specific knowledge expert.
It based on the mode that regular expression extracts, is handled using web document as character stream.One is write out by user
Kind of mode matches the information that user wants.Information regards character stream above-mentioned as.In the treatment process to information
In, it is matched to the information to be extracted rapidly by using regular expression.The advantages of carrying out information extraction using regular expression
It is that there is very high efficiency and accuracy to the information extraction of known features, does not need a large amount of artificial mark sample,
Advanced field priori knowledge is not needed.After work order classification, every one kind work order generally has fixed expression pattern, is based on canonical
The information extraction method of expression formula is relatively applicable in this scene.This research is based on regular expression and defines several information pumping
Modulus plate realizes the accurate extraction of the information such as failure system title, IP, physical location, phenomenon of the failure.
Further, the convergence work order and the not converged work order include at least in terms of following two: one or more
A failure is in one or more one or more work orders for monitoring and generating under dimension；One or more caused by one or more reasons
The appearance repeatedly that a alarm phenomenon is inscribed in one or more.It includes two classes that work order, which can be restrained: first is that a certain failure is in difference
The multiple work orders generated under monitoring dimension, the switching of the virtual machine as caused by server failure, it is possible to create port is obstructed, the page is beaten
It does not open, the different alarms such as process number is abnormal.Second is that repeatedly appearance of the alarm phenomenon caused by a certain reason in specific time, such as certain
The excessively high alarm of CPU usage caused by one batch processing job, may continue to occur in morning for more days.
In the work order convergence model of same event, input includes alarm time, alarm event type, systematic name, collection
Group's title, IP etc., input feature vector dimensional comparison is small, and the methods of deep neural network is not applicable.The present invention uses decision Tree algorithms
Realize that work order restrains model.It is random gloomy using being formed including more decision trees to prevent over-fitting of the model on sample
Woods model.
The information such as fault category, systematic name, time of origin, the IP obtained according to preceding step are believed in conjunction with CMIS system
Breath, first temporally clusters work order, then merges work order according to topological relation, affiliated group, fault type etc..This convergence
The polymerization work order that method generates, the different manifestations of usually same failure.For same class failure, the rule of time of origin is excavated
Property.Such as: the similar failure occurred repeatedly in the same time prompts us the basic reason of failure not solve, needs into one
Walk analyzing and positioning.This research has navigated to several recurrent alarm events, disappears for subsequent reason investigation and hidden fault
Except providing foundation.
Further, the input of root cause analysis model is the polymerization work order after convergence, and input feature vector includes alarm event class
Type, systematic name, cluster name, IP, characteristic dimension is smaller, and the methods of deep neural network is not applicable.This research uses decision
Tree algorithm realizes that work order restrains model.To prevent over-fitting of the model on sample, using what is formed including more decision trees
Random Forest model.
Fig. 4 shows another embodiment of the present invention, and one kind realizing the convergent control of work order based on feature modeling
The module connection diagram of device processed, comprising: the first determining device 1: work order determines one or more works based on one or more
The working principle of one or more characteristic informations in list, first determining device 1 can refer to abovementioned steps S101, herein
It will not go into details.
Further, the control device further includes the first acquisition device 2: one by one by one in one or more work orders
Or multiple characteristic informations characteristic model corresponding with the characteristic information is matched, and one or more convergence work orders are obtained
And/or one or more not converged work orders, the working principle of first acquisition device 2 can refer to abovementioned steps S102,
It will not go into details for this.
Further, the control device further includes the first processing unit 3: matching and one or more convergence work orders and/
Or the work order information that one or more not converged work orders are adaptable, and the work order information is subjected to Dynamic Display, described first
The working principle of processing unit 3 can refer to abovementioned steps S103, and it will not be described here.
Further, first acquisition device further includes the first judgment means 21: the work order classification based on the work order
Information judges whether to converge to fisrt feature model group, and the working principle of first judgment means 21 can refer to aforementioned step
Rapid S1021, it will not be described here.
Further, first acquisition device further includes the second judgment means 22: in fisrt feature model group
In the information extraction information based on the work order judge whether to converge to second feature model group, second judgment means 22
Working principle can refer to abovementioned steps S1022, it will not be described here.
Further, first acquisition device further includes third judgment means 23: in second feature model group
In the work order coverage information based on the work order judge whether to converge to third feature model group, the third judgment means 23
Working principle can refer to abovementioned steps S1023, it will not be described here.
Further, first acquisition device further includes the 4th judgment means 24: in third feature model group
In the root cause analysis information based on the work order judge whether to converge to fourth feature model group, the 4th judgment means 24
Working principle can refer to abovementioned steps S1024, it will not be described here.
It will be appreciated by those skilled in the art that the invention mainly comprises off-line model training cluster and production application cluster two
Point, the intellectual product of each scene passes through this framework and completes life cycle management.Number is mainly completed in off-line training model cluster
It is distributed to production according to processes, the model after being verified such as acquisition and pretreatment, feature selecting, model training, verifying publication and answers
With cluster, interface is provided upwards and is embedded in other application system, is associated with acquisition rule with the knowledge platform of professional knowledge storage downwards
Scheme.Offline and production application cluster is managed by container platform is included in.Training framework constructs a closed loop path.Day
Normal real-time running data enters formal model and carries out machine mark, and is compared with the artificial annotation results of each interface platform, poor
Different result differentiates that recognition result is trained again into training pattern into secondary expert, so recycles, and makes model identification judgement
Effect is more acurrate.The research process of the present invention perfect application platform management of intelligent O&M simultaneously, construct one it is complete
Model life cycle path of Research contains the upgrade services such as a variety of machine learning algorithms, model training visualization, is more intelligence
Energy O&M scenarios provides procedure and exempts from the AI implementing ability intervened.
For root because of convergence, algorithmic procedure uses LSTM, and ((Long Short Term network) is to the failure classes of work order
Type is classified, and using segmenting in natural language processing, part-of-speech tagging combination regular expression carries out the extraction of key message.
Using the Random Forest model for including more decision trees composition, to warning information in time, alarm type, interconnected system, pass
Connection cluster, association IP are associated item excavation, in combination with system topological relation information in CIMS (configuration information management system)
It merges.
Fig. 5 shows the third embodiment of the present invention, and one kind realizing the convergent control device of work order based on feature modeling
Module connection diagram, comprising: the second acquisition device 4: based on one or more data acquisition modes in one or more numbers
One or more original work order informations are obtained according to acquisition platform, it will be appreciated by those skilled in the art that the work of the second acquisition device 4 is former
Reason can refer to abovementioned steps S201, and it will not be described here.
Further, the control device further includes the second determining device 5: to one or more of original work order informations
It is pre-processed, one or more pretreatment work order informations is determined, before the working principle of second determining device 5 can refer to
Step S202 is stated, it will not be described here.
Further, the control device further includes third determining device 6: being believed one or more of pretreatment work orders
Breath carries out feature extraction, determines one or more work order characteristic information, and the working principle of the third determining device 6 can refer to
Abovementioned steps S203, it will not be described here.
Further, the control device further includes the 4th determining device 7: being believed based on one or more of work order features
Breath modeling determines one or more characteristic models to match with the work order characteristic information, the work of the 4th determining device 7
Making principle can be with reference to abovementioned steps S204, and it will not be described here.
Specific embodiments of the present invention are described above.It is to be appreciated that the invention is not limited to above-mentioned
Particular implementation, those skilled in the art can make various deformations or amendments within the scope of the claims, this not shadow
Ring substantive content of the invention.
Claims (13)
1. one kind realizes the convergent control method of work order based on feature modeling, based on dynamic acquisition data, rule modeling is summarized,
And work order data application is realized to work order restrains in modeling, which comprises the steps of:
A. work order determines one or more characteristic informations in one or more of work orders based on one or more；
B. one by one by one or more characteristic informations character modules corresponding with the characteristic information in one or more work orders
Type is matched, and one or more convergence work orders and/or one or more not converged work orders are obtained.
2. control method according to claim 1, which is characterized in that further include step c after the step b: matching
The adaptable work order information with one or more convergence work orders and/or one or more not converged work orders, and the work order is believed
Breath carries out Dynamic Display.
3. control method according to claim 1 or 2, which is characterized in that in the step a, the characteristic information is extremely
Include: less
Work order classification information；
Information extraction information；
Work order coverage information；And
Root cause analysis information.
4. control method according to claim 3, which is characterized in that the step b includes the following steps:
B1. the work order classification information based on the work order judges whether to converge to fisrt feature model group, if so, executing step
Rapid b2；
B2. the information extraction information in fisrt feature model group based on the work order judges whether to converge to the second spy
Model group is levied, if so, executing b3；
B3. the work order coverage information in second feature model group based on the work order judges whether to converge to third spy
Model group is levied, if so, executing b4；
B4. the root cause analysis information in third feature model group based on the work order judges whether to converge to the 4th spy
Levy model group.
5. control method according to claim 1 or 2 or 4, which is characterized in that the foundation of the characteristic model will be by such as
Under type obtains:
I: one or more original works are obtained in one or more data acquisition platforms based on one or more data acquisition modes
Single information；
Ii: pre-processing one or more of original work order informations, determines one or more pretreatment work order informations；
Iii: carrying out feature extraction to one or more of pretreatment work order informations, determines one or more work order feature letter
Breath；
Iv: being based on one or more of work order Feature information modelings, determines one or more and the work order characteristic information phase
Matched characteristic model.
6. control method according to claim 5, which is characterized in that the data acquisition modes include at least such as under type
Any one of or appoint it is a variety of:
SQL export；
Third party API export；Or
Crawler acquisition.
7. control method according to claim 5, which is characterized in that the data acquisition platform includes at least such as lower platform
Any one of or appoint it is a variety of:
O&M worksheet processing system；
Configuration management database；
Application performance monitoring；
Banner system；Or
Unified log platform.
8. control method according to claim 5, which is characterized in that the original work order information includes at least following information
Any one of or appoint it is a variety of:
Work order；
CMDB；Or
APM。
9. control method according to claim 5, which is characterized in that the pretreatment includes attachment acquisition, data screening
And data encoding.
10. control method according to claim 1 to 9, which is characterized in that the convergence work order and described
Not converged work order includes at least following two aspects:
One or more failures are in one or more one or more work orders for monitoring and generating under dimension；
The appearance repeatedly that the alarm of one or more caused by one or more reasons phenomenon is inscribed in one or more.
11. one kind realizes the convergent control device of work order based on feature modeling characterized by comprising
First determining device (1): work order determines one or more features letter in one or more work orders based on one or more
Breath；
First acquisition device (2): one by one by the one or more characteristic informations and the characteristic information in one or more work orders
Corresponding characteristic model is matched, and one or more convergence work orders and/or one or more not converged work orders are obtained.
First processing unit (3): matching is adapted with one or more convergence work orders and/or one or more not converged work orders
Work order information, and the work order information is subjected to Dynamic Display.
12. control device according to claim 11, which is characterized in that first acquisition device includes the following:
First judgment means (21): the work order classification information based on the work order judges whether to converge to fisrt feature Model Group
Group；
Second judgment means (22): in fisrt feature model group based on the work order information extraction information judgement be
It is no to converge to second feature model group；
Third judgment means (23): in second feature model group based on the work order work order coverage information judgement be
It is no to converge to third feature model group；
4th judgment means (24): in third feature model group based on the work order root cause analysis information judgement be
It is no to converge to fourth feature model group.
13. control device according to claim 11, which is characterized in that further include:
Second acquisition device (4): one is obtained in one or more data acquisition platforms based on one or more data acquisition modes
A or multiple original work order informations；
Second determining device (5): pre-processing one or more of original work order informations, determines one or more pre- places
Manage work order information；
Third determining device (6): carrying out feature extraction to one or more of pretreatment work order informations, determines one or more
Work order characteristic information；
4th determining device (7): being based on one or more of work order Feature information modelings, determines the one or more and work
The characteristic model that single characteristic information matches.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910423511.7A CN110175272A (en) | 2019-05-21 | 2019-05-21 | One kind realizing the convergent control method of work order and control device based on feature modeling |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910423511.7A CN110175272A (en) | 2019-05-21 | 2019-05-21 | One kind realizing the convergent control method of work order and control device based on feature modeling |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110175272A true CN110175272A (en) | 2019-08-27 |
Family
ID=67691614
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910423511.7A Pending CN110175272A (en) | 2019-05-21 | 2019-05-21 | One kind realizing the convergent control method of work order and control device based on feature modeling |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110175272A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111178537A (en) * | 2019-12-09 | 2020-05-19 | 华为技术有限公司 | Feature extraction model training method and device |
| CN112183782A (en) * | 2020-10-13 | 2021-01-05 | 中国联合网络通信集团有限公司 | Fault work order processing method and equipment |
| CN113542039A (en) * | 2021-09-16 | 2021-10-22 | 浩鲸云计算科技股份有限公司 | Method for positioning 5G network virtualization cross-layer problem through AI algorithm |
| CN115695142A (en) * | 2022-10-25 | 2023-02-03 | 浪潮通信信息系统有限公司 | Network operation and maintenance oriented event monitoring method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1980444A (en) * | 2005-12-02 | 2007-06-13 | 上海移动通信有限责任公司 | Subscriber attaching registor data inquiring and processing concentrated convergent integrating method |
| CN109165808A (en) * | 2018-07-18 | 2019-01-08 | 南瑞集团有限公司 | A kind of power telecom network scene work order distributing method |
-
2019
- 2019-05-21 CN CN201910423511.7A patent/CN110175272A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1980444A (en) * | 2005-12-02 | 2007-06-13 | 上海移动通信有限责任公司 | Subscriber attaching registor data inquiring and processing concentrated convergent integrating method |
| CN109165808A (en) * | 2018-07-18 | 2019-01-08 | 南瑞集团有限公司 | A kind of power telecom network scene work order distributing method |
Non-Patent Citations (1)
| Title |
|---|
| 胡罡: "https://e.huawei.com/cn/publications/cn/ict_insights/201809301120/case/201810080931", 《华为期刊》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111178537A (en) * | 2019-12-09 | 2020-05-19 | 华为技术有限公司 | Feature extraction model training method and device |
| CN111178537B (en) * | 2019-12-09 | 2023-11-17 | 华为云计算技术有限公司 | Feature extraction model training method and device |
| CN112183782A (en) * | 2020-10-13 | 2021-01-05 | 中国联合网络通信集团有限公司 | Fault work order processing method and equipment |
| CN112183782B (en) * | 2020-10-13 | 2024-04-12 | 中国联合网络通信集团有限公司 | Fault work order processing method and equipment |
| CN113542039A (en) * | 2021-09-16 | 2021-10-22 | 浩鲸云计算科技股份有限公司 | Method for positioning 5G network virtualization cross-layer problem through AI algorithm |
| CN115695142A (en) * | 2022-10-25 | 2023-02-03 | 浪潮通信信息系统有限公司 | Network operation and maintenance oriented event monitoring method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110175272A (en) | One kind realizing the convergent control method of work order and control device based on feature modeling | |
| CN107861942B (en) | Suspected power complaint work order identification method based on deep learning | |
| CN111240662B (en) | Spark machine learning system and method based on task visual drag | |
| CN113094200B (en) | Application program fault prediction method and device | |
| CN110942387B (en) | Method and system for constructing electric ticket business function based on micro-service | |
| CN109871401A (en) | A kind of time series method for detecting abnormality and device | |
| CN108021929A (en) | Mobile terminal electric business user based on big data, which draws a portrait, to establish and analysis method and system | |
| CN107317724A (en) | Data collecting system and method based on cloud computing technology | |
| US11416754B1 (en) | Automated cloud data and technology solution delivery using machine learning and artificial intelligence modeling | |
| CN110110095A (en) | A kind of power command text matching technique based on shot and long term memory Recognition with Recurrent Neural Network | |
| CN111368926B (en) | Image screening method, device and computer readable storage medium | |
| CN105930347A (en) | Text analysis based power outage cause recognition system | |
| CN108985467A (en) | Secondary device lean management-control method based on artificial intelligence | |
| CN110308946A (en) | Race batch processing method, equipment, storage medium and device based on artificial intelligence | |
| CN114757307B (en) | Artificial intelligence automatic training method, system, device and storage medium | |
| US20210141779A1 (en) | System and method for facilitating an objective-oriented data structure and an objective via the data structure | |
| EP3785128A2 (en) | System and method for creating recommendation of splitting and merging microservice | |
| CN115567370B (en) | Automatic alarm method, storage medium and system based on real-time event data | |
| US20220291966A1 (en) | Systems and methods for process mining using unsupervised learning and for automating orchestration of workflows | |
| CN113947468B (en) | Data management method and platform | |
| CN109635089B (en) | Literature work novelty evaluation system and method based on semantic network | |
| CN116010187A (en) | Log detection method and related device | |
| CN113312924A (en) | Risk rule classification method and device based on NLP high-precision analysis label | |
| CN116739408A (en) | Power grid dispatching safety monitoring method and system based on data tag and electronic equipment | |
| CN110086166B (en) | Method and system for representing stable operation quota of power grid |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190827 |